Definitely, Maybe Agile
Definitely, Maybe Agile
Five AI Predictions for 2026
As we close out 2025, Peter and Dave are making predictions about what's coming in 2026, especially around AI, organizational change, and how teams actually work.
They cover five key predictions:
- AI moves from tools to organizational capability: Organizations that invest in literacy, governance, and data foundations will pull ahead of those just sprinkling AI on top and hoping for the best.
- Critical thinking beats prompt engineering: The real competitive advantage won't be writing clever prompts. It'll be knowing when to pause, think through the problem, and decide if you even need the AI in the first place.
- Product delivery becomes non-negotiable: After 20 years of pushing Agile principles, AI might finally force organizations to actually adopt them (even if they're reluctant to call it "Agile").
- Businesses return to fundamentals: Just like the dot-com bubble, we're heading toward a moment where the market will care more about revenue, customers, and sustainability than hype.
- Reskilling becomes a structural investment: Organizations will need to figure out what roles actually look like in an AI-enabled world and invest in growing their people, not just replacing them.
At the end, Peter and Dave pick which prediction is hardest to measure (spoiler: it's critical thinking) and commit to revisiting these in March to see how wrong they were.
If you've been wondering where all this AI stuff is actually heading, this episode cuts through the noise with grounded, practical predictions you can actually use.
Related episodes:
- AI and Knowledge Management with Derek Crager: https://www.buzzsprout.com/1643821/episodes/17360635
- Product vs. Process Innovation: https://www.buzzsprout.com/1643821/episodes/7953100
- There Are No Safe Bets in Business Anymore: https://www.buzzsprout.com/1643821/episodes/17433034
Reach out: feedback@definitelymaybeagile.com
Peter: 0:04 Welcome to Definitely Maybe Agile, the podcast where Peter Maddison and David Sharrock discuss the complexities of adopting new ways of working at scale. Hello, Dave.
Dave: 0:14 How are you today? Peter, great to catch up with you. Great to just have one of those chats again. We've had a lot of guests over the last few weeks, which has been wonderful. There's some fantastic conversations we've had. I think this is one of our chances just to take a pause, look around us, and maybe make a few predictions rather than looking back. I know it's the end of a year, but instead of looking back, looking forward and just catching up a little bit on what we're seeing around us.
Peter: 0:41 Yeah, this is going to be one of those chats about this moment in time, so that we can look back a year from now and be proved completely wrong.
Dave: 0:57 We'll know by March.
Peter: 0:59 Yes, yes. So maybe we should look back in March at these predictions and see how true they were. Would you like to kick us off? We have five predictions for the audience, right?
Dave: 1:09 So the idea for today is five predictions that we see around us for next year. Out of those five, they're nearly all going to say something about AI. I think in digital product delivery, everybody is talking about AI. Organizational change, digital transformations, however you name it or call it, people are talking about AI. So maybe if we just kick things off, one of the things that we are seeing is a lot of organizations trying to get onto the AI bandwagon, but not really landing. I think one of the key things is going to be this push away from standalone little proofs of concept around trying out different tools and things like this, but really investing in getting an organizational capability around AI. The only investment that I see certainly in the organizational capability is, hey, you've got Copilot or name your large language model of preference that you've bought licenses for, that's now allowed to be used in the work environment, but it's very little in the way of explaining how to embed that in the way you do the work, how to actually get organizations empowered to leverage the various tools that are made available across an organization.
Peter: 2:34 And I think this is a critical piece. If we're predicting something, we're predicting that organizations that do invest in providing the literacy, governance, and data foundation pieces to their organization are going to see greater success with the AI tools and building that capability within the organization than the ones who effectively try to sprinkle it on top of the organization and hope that, well, it's everybody's got it, it's ubiquitous, everyone knows how to use it, so they must understand how to go about building things with this if we just give it to them. Because they're all asking for it, right? All of our staff are asking for it.
Dave: 3:16 I just wanted to say, Peter, you touched on a really critical piece of that, which is the regulatory foundation on which things are built. Because compliance is really beginning to, and security, is one of the reasons why executives are cautious. They're standing back a little bit or they're a little bit worried about pushing the accelerator down. So one of the things, and you just touched on it around governance and data security and just understanding how to configure and set things up and make sure you're protecting the assets that your organization has, making sure security and data protection is in place. Those basics are often either not clearly understood or articulated, or just barely in place and maybe not fully instrumented so that we know that they're in place and we can move with confidence. That's a critical piece coupled with literacy from an organizational perspective. And the thing that bugs me about that one, it isn't about prompt writing. It's not about this prompt engineering thing, it has nothing to do with that, it has to do with where can I use these tools in my area, where should I not? Because people are going to get burned, they're going to write the wrong email, and we're seeing this all over, right? Organizations try and use it in customer service, and that causes these things that hit the news. How do we remove the fear of working in the wrong place?
Peter: 4:45 And I see a couple of things commonly happening there. One is, okay, if we can generate a lot more text, which then either has to get summarized for the humans on the other end to understand it, or you've basically created so much more noise that all of your people are now just so much more busy trying to figure out what's going on. The concept I think that's bandied in the media around work slop is a part of it. This idea that there is just lots of, it's very easy to generate content. We've still got the same number of people trying to absorb that content. Now, the only way that necessarily works is, well, instead of the people absorbing that content, we have AIs absorbing that content. Well, at what point does that cycle?
Dave: 5:29 Yeah. I actually want to move on to a second prediction because I think this is a really great one, which is that everybody's talking about everybody's jobs being replaced by AI. But at the very least, we're in this middle period where not everything is currently going to get replaced with AI, it's going to happen soon enough. But there's this sort of, the people who blindly use AI who just end up throwing, you know, you can see them in meetings where they're literally throwing everything into a prompt and then generating outcomes in real time with conversations, for example. That's one of the things that we see. But what I really look at there is the importance of critical thinking. It's actually the pause before you draw on the large language models you're using. What is my expected outcome, how am I going to be able to interpret what comes back in terms of whether it's good or whether it's bad. So that critical thinking being more valuable than being an amazing prompt engineer.
Peter: 6:34 Yeah, and I think it comes back to what I was saying about, like, if I've got a lot of text to review and all of this to understand, it becomes harder necessarily to make that decision. And if I then start to fall back and rely entirely on the AI to make those decisions before I've thought through what it is I'm trying to achieve, then potentially I find myself in a situation where I'm not achieving the outcome that I wanted to. So having the ability to think critically about what is it I'm trying to do, what are the trade-offs I'm going to make, what is the direction I need to go in. Have I thought through that before immediately going to the LLM and relying entirely on it to come up with the solution? I think we're going to see more emphasis on encouraging that in organizations. We do see it a little in, I mean the space I operate in, the concepts around human in the loop. We want to make sure that somebody's actually always there looking at this. I think we haven't solved the problem of how busy everybody is to actually have the time to step back or to think about how roles are changing as a consequence of this. So, but as the role changes, you need to operate in a different way. And that means you've got to have the space to do so. And if you're still trying to operate in the old way and now you're trying to operate in this new way at the same time, that's where things will fall through the cracks.
Dave: 8:09 And I don't know, there's a paper that I've heard referenced around, like, the research focused on people who went to large language models to make their decisions for them. And they've very rapidly lost the ability to be able to make decisions. And I think this is the danger we get into, is why critical thinking becomes so important. Organizations don't need people who enter prompts into a large language model to get responses. They need people who are thinking about the problems and using the tools around them to rapidly reach some sort of workable, actionable solution. But actually the critical thinking still becomes the differentiator. In fact, it strengthens the fact that that is happening. I think it's almost human before the loop. It's both human in the loop, but human before the loop.
Peter: 9:00 I think that takes us to the third prediction around product delivery. As we introduce these tools and these capabilities into organizations, what we've seen as the concept of agile suddenly becomes a little bit more not only needed but required, but we don't really want to call it agile anymore because that ship has sailed. It's more about having that understanding of are we building the thing that's going to deliver value? Do we have the right pieces in place to understand that this is the correct solution? And that's where we will start to see more reliance, I think, on, again, that changing of roles into product owners with a much richer source of data and understanding, being able to understand and make judgment calls as to is this the right thing? Are we building the right solution for the market? Are we doing the right pieces? Because we can now start to understand that value much more rapidly.
Dave: 10:08 I love this one, Peter, because both of us have been talking about that sort of existential crisis around agile. It's no longer, it's so important still, but it's in the foundations of everything. Everybody kind of goes, we're already agile, and they don't focus on it. And yet at the same time, a lot of the state of AI, you know, different reports that are coming out from lots and lots of different places, are all touching on the fact that if you don't know why you're building something, if you don't kind of watch that whole product-centric mindset, if you don't have that right, you're going to have a headache. And I'm reminded of SAFe when Scaled Agile Framework came out. When you read it, it said the assumption is you already have functioning agile and Kanban delivery teams in place, of which then you could then build this whole scaling framework around the sides of it. This is the same. The assumption is as you roll out AI, you're focused on the fact you can get ideas out of the door, and those ideas are aligned with what your customers are wanting, and you know how to do it.
Peter: 11:17 Yeah, which is, it's fascinating. I find that there seems to be this reluctance to call it agile, but essentially the delivery model is exactly what Agile had been preaching, for lack of a better word, for the last couple of decades. So I'm going to get my collar somewhere. Yes. So yeah, I think it's very interesting how that's sort of coming about. I think in some ways you could say that the introduction of AI is driving the kind of transformation that agile practitioners have been pushing for for quite a long time.
Dave: 11:59 Which is great. It means we'll hopefully be busy next year. You think. The next one really, I was just thinking about this. Everybody's talking about AI bubbles, whether or not the bubble is there, the AI hype and things like this. And again, people of a certain age will remember the similar sort of conversations going on around the internet and around 2000 and whether the internet was a bubble or not. And the interesting thing that came out of that is this shift away from any idea is a good idea, it just needs to move to become revenue generating, but we're not worried about that right now. And after the internet sort of boom and bust cycle went through, all of the focus was on businesses and their fundamentals. Did they have customers? Were those customers buying? Were they returning? Could you make money on what you were selling your product for? All of the things that we think of as how to operate a business. And I think this is something that we're going to see coming in next year as we really start focusing on those business metrics that accountants care about.
Peter: 13:13 Yeah, I think we're going to start to see that in 26 for sure, because it becomes very difficult to actually differentiate between some of these solutions out there. When you can so rapidly iterate and put solutions into the market, they very easily can look at their neighbor and go, oh, look, I can do that. And so the ability to differentiate and the number and the proliferation of solutions out there means that not everybody's going to survive. I think there is definitely, I mean, AI is certainly transformational. I don't think we could say for a second that the hype cycle would say that everybody, all of these companies are going to fail, not at all. It is going to change a lot of operating models in organizations and a lot of the way that we do things. But I agree and almost hope that we get back to a focus on what is the business intending to do. And along with that, what I would hope to happen, as per our previous predictions and understanding those fundamentals, is getting to the point of thinking about how can we use this technology to fundamentally change the way our business processes operate, versus just replacing parts of those business processes with a non-deterministic agent, which isn't, because that doesn't necessarily move us forward.
Dave: 14:41 Well, exactly. It's sort of instead of calling everything an AI problem that needs an AI solution, it's beginning to use that cognitive thinking that we just talked about, cognitive problem solving, or the explicit looking at how to think of a problem, critical thinking, to identify non-deterministic versus deterministic problems. Deterministic problems, go and automate them. That's a different thing, right? And then using that understanding, the kind of thing, the problems that you're looking at to then actually not have a promise that says in three years' time we're going to suddenly hit some sort of a gold mine of revenue, but how do we build out a new opportunity in exactly the same way you'd expect, you know, early wins, incrementally delivering value, building out a customer base, a loyalty based on those early learnings as you go through those basics around customer value, around flow, around financial outcomes, things like that.
Peter: 15:41 Exactly. Yeah. Prediction number five.
Dave: 15:44 Last one, right?
Peter: 15:45 Yes. Do you want to kick us off on this one?
Dave: 15:48 Right. So this one is, I think this one's probably the hardest one for us to pin down. So the other ones feel like they're almost already emerging and happening right now. This one is about the structural investment into employee upskilling and reskilling and education that goes there. We're coming out of a scenario where there's less spending on training and education, and yet there's never been a greater need to redefine your role or my role in terms of how we create value for the organization. And that one is either going to, there's going to be something, I'm pretty sure, around this focus on structural investments into reskilling or upskilling of your employee base. And I just, yeah, there will be organizations that just kind of move people out and move new people coming in, but that assumes you've got a market of people who have the right skills. And I think there's a lot of skills which are still not well established that we're going to have to explore and build out. And there's a lot of value in just having some very capable individuals in your organization who can kind of generate, you know, emergent opportunities and make those work. It's just difficult to exactly put a finger on exactly how that's going to work.
Peter: 17:12 And I think some of the difficulty there is that we're not a hundred percent clear on what the end goal is. Like, what is that end role that you're reskilling to? Like what is the set of roles that you need in an organization for effective delivery of solutions? We might have general titles, but what are the skills that we're going to need tomorrow? So I think, I agree there's a, I think the prediction here is the creation of that understanding of, like, what is the new team look like. Like if we look back and we think of the two pizza team and what that makeup would look like, which would be dependent partly on the underlying system architecture, but we could pretty much say, okay, I need a team that looks like this. Like it's going to have this many developers, and they'll have some of the things that we're going to do.
Dave: 18:06 We talked about the replacement of two pizza teams with one pizza teams supported by various AI agents and tools and so on. And there's a need for, I mean, we talked about it in our conversation today about just things like critical thinking, things like not just being prompt engineering, but actually knowing a little bit more about how to use these tools. All of these require skills which a lot of the ingredients are there, but they really need some sort of a definition of some career progression or a ladder of learning that needs to be put in place to see that happen. And the interesting thing is it becomes very contextual. Your organization will need a different skill set to my organization, and therefore we can't just go into a market and get some general diploma or certificate around it. It's something that we probably need to look at from ourselves because there's strategic value in knowing exactly what learning we're creating in our organization.
Peter: 19:03 And I'm reminded somewhat of the product owner role, and, like, how do you train somebody to be a product owner? Well, there's a myriad of skills that you need, but there are good product owners and bad product owners, and we can train you on certain aspects of this, but, like, where? So I think putting together those pathways, understanding, like, what we consider to be a good set of skills for somebody in a delivery team, and how many people do we need in those teams and what does that look like? I can see that becoming a topic of conversation over the next sort of three, six months.
Dave: 19:46 So should we summarize, wrap things up?
Peter: 19:49 Sure.
Dave: 19:49 So then maybe I'm going to ask you which one of these gets blown up first.
Peter: 19:54 You mean which one of these doesn't happen?
Dave: 19:56 Doesn't happen, yes. First, because I'm sure many of them will not.
Peter: 20:01 So I will quickly run through and I'll run through with the titles that we had here. So prediction one was AI moves from tools to organizational capabilities. So the concepts around we need to have literacy, governance, data foundations in place so that we can move beyond short-lived pilots that never make it to production. The second prediction was around product delivery becomes non-negotiable. We start to do the agile things that we've been asking people to do for 20 years. They actually now become a part of how we operate because we can. I think some of it is we can actually, the organizations can more rapidly actually see the value from it than they could before. Prediction three was critical thinking outpaces prompting as a real competitive advantage. So having smart people who can actually use their judgment to understand whether what they're being told by the LLMs is actually valuable or not, or that they can actually think through the problem, work out what it is before they even go and ask. So critical thinking skills, judgment. Prediction four, businesses returning to fundamentals after the AI hype cycle. So we were talking about this around how at the moment it seems that everybody and their buddy can spin up an AI company and get millions of dollars in investment despite having no customers or any possible hope of having a revenue stream. But hey, let's move forward. The next one, prediction five. So we're hoping to see an end to that, not an end to it, but at least a slowing down of that. Reskilling is a structural investment, not a perk. The necessity to understand what are the roles that you're going to need and how am I evolving my workforce and how do I get that into place? How do I figure out how to grow the people that I have? I think that is going to, I know a lot of people I'm working with are talking about this at the moment and trying to figure out what that is so they can put those plans into place next year. And so I could predict that that is going to become a broader conversation for sure. I'll ask a question to you.
Dave: 22:15 Which of these do you think is going to, so, I loved listening to you reading through those because the one that strikes me as probably the hardest one to see coming true is prediction three, critical thinking outpacing prompting or becoming a real competitive advantage. My reasoning here is it's pretty invisible. I think it's really difficult to tell if your organization really is brilliant at critical thinking and the feedback loop is a bit too long for us to be able to just say, yeah, it's working or it's not working. So as much as I think it is going to be the superpower, I don't know how organizations will actually make that a measurable thing that they can get really good at. I think we'll just, you know, hindsight is 2020, we're going to look back and say they were really good at critical thinking without really understanding how they made that happen.
Peter: 23:10 Yeah, I agree with you there. I think it is going to be the hardest one to measure. It's, we recognize it when we see it.
Dave: 23:21 Exactly.
Peter: 23:22 But it's a very hard thing to teach people to. Yeah. So I can see that as being a difficult one moving forward. So yeah, I think I might pick that one too, actually, as out of the five. All right, well, pleasure, Dave, as always. That's always fun. Yeah, so I hope this was valuable to our listeners and you can reach out to us at feedback@definitelymaybeagile.com if you have any questions. Over the holiday season, we'll be re-releasing some of our earlier episodes, the most popular ones, the ones that get downloaded the most. And hope you enjoy those and look forward to talking to folks in the new year.
Dave: 24:05 Yeah, I'm kind of looking forward to next year because there might be a hint of a rebrand in the offing as well. We'll have to think about that and see what works.
Peter: 24:13 Yes, yes.
Dave: 24:15 All right, talk to you soon.
Peter: 24:16 Talk soon. You've been listening to Definitely Maybe Agile, the podcast where your hosts Peter Maddison and David Sharrock focus on the art and science of digital, agile, and DevOps at scale.
.png)
.pdf.png)